During the development of any new medical device, various tests may be required, including the characterization of physical properties (geometric, mechanical, electrical, electromagnetic, thermal, chemical, etc), the evaluation of overall device performance (numerical simulation or simulated use testing), or testing to determine the effect of the device on living tissues. These development tests may be broadly classified as either biological tests, theoretical tests, or physical tests, although there are areas where these testing classes overlap one another.
Biological testing generally involves an analysis of the interaction between the device and human or animal tissues. The biological tests that are performed first are generally biocompatibility tests, which evaluate the tendency of the device to cause damage to living tissues by mere presence of the materials comprising the device. Later on in the development cycle, the device may be tested in a live animal (animal study) or a human patient (clinical trial) to determine the ability of the device to perform its intended use and to evaluate safety and efficacy (device performance). Animal studies represent a special type of test known as simulated use testing, so called because the animal is a simulation of the actual use (human) environment.
Theoretical or computational tests may include finite element analysis, kinematic analysis, and computational fluid dynamics. These tests employ knowledge of the physical properties (strength, mass, density, viscosity, etc) of the device and actual use environment to construct a computer model of the device—tissue system. This type of model may then be used to predict device performance, the tendency of the device to fail, and possibly the tendency of the device to cause injury. Of course, these models are limited by the assumptions made in their derivation and the computational power of the computer. Unfortunately, it may be difficult to quantitatively describe a device, and more importantly the actual use environment, in sufficient detail to yield realistic results.
Physical testing essentially evaluates the design of the device. That is, this physical testing may involve; (1) the measurement of device geometry such as lengths, diameters, and wall thicknesses, (2) the measurement of mechanical properties such as tensile strength and stiffness, (3) the measurement of other device characteristics such as color, thermal conductivity, dielectric properties or other properties, or (4) simulation testing involving trial use of the device in some model of the actual use environment. The purpose of this simulation testing is to evaluate the safety (tendency to injure) and efficacy (performance characteristics) of the device, and in general to evaluate the ability of the device to perform it's intended use. As previously stated, animal studies are one important form of simulation test. Other vehicles (the simulated environment) for this type of testing include cadavers (both human and animal) and benchtop fixtures, which are man-made representations of a particular target anatomy.
The new FDA quality system regulation (QSR) now requires testing under simulated or actual use conditions for all nonexempt Class II and Class III medical devices. Not all manufacturers perform actual use (human clinical trial) testing for every medical device, so in these cases simulation testing is definitely a requirement. At least four traditional simulation options are available to meet this requirement, each with its own advantages and drawbacks. These four general approaches (Table I) to simulation testing involve theoretical (computer) models, benchtop (physical) models, cadaver (human or animal) models, and live animal models. Once again, human subjects are also employed in the development of many medical devices, but since humans represent the actual use environment, these tests (clinical studies) are not considered simulation tests.
TABLE IThe four general approaches to simulation testingincluding the environment and models involved.BenchtopCadaverLiveTheoretical ModelModelModelAnimalIn vitroIn vitroIn vitroIn vivoTheoretical orPhysicalHuman orAnimalcomputational modelpropertiesanimalstudymodelcadaver
Typical medical device development schemes generally involve testing early prototypes in simple bench top test fixtures. Feedback from these tests shape the product through design revisions that are subsequently evaluated using the same model. However, since this process is iterative, as the design matures the models that are needed generally become more complex. For example, a new coronary catheter may undergo initial testing in simple plastic tubes, followed by glass models designed to mimic the size and geometry of the coronary vasculature. The product may experience a series of changes resulting from these tests until the designer is satisfied with performance, and once a certain level of confidence is achieved the testing will proceed to the next available model. In the medical device industry this model is generally a live animal.
In practice, the medical device industry typically employs one or more of the four previously mentioned (Table I) model types in simulation testing prior to seeking approval for human use (a clinical trial). Of course, common sense dictates that the model selected be representative of actual use conditions, but only the clinical trial, which is not a simulation test, fully satisfies this criteria. Unfortunately, human subjects are unavailable for use until late in the development cycle due to risk, regulatory, and ethical considerations. A live animal model has therefore traditionally been the next best choice.
Animal models are currently the gold standard of pre-clinical trial medical device simulated use testing. In fact, the quality of data produced in these studies can be very high, particularly if the proper animal model is selected, the device and protocol are well designed, and the correct number of animals is used. Designed experiments are possible and are commonly employed, but require an increase in the number of animals. These tests are also performed under physiological (for the animal) conditions. Unfortunately, these studies are expensive because of the staff and facilities required to support the work. A registered facility must be contracted to run the study and care for any animals purchased, a surgeon must be retained to perform the required procedures and to generate the study protocol, and the services of a veterinarian, anesthesiologist, and surgical aide are also required. These studies can easily exceed $100,000 in total costs, and grow even more costly as the number of animals is increased.
The inability to test prototype devices on human subjects is the reason medical device developers resort to animal studies in the first place. Still, animal models suffer from a whole range of unique problems, including the many deviations between human and animal anatomy and physiology, the confounding effects of variation between individual animals, and the unpredictability that arises from using a model that is extraordinarily complex.
Animal models may include live canine, porcine, or bovine specimens, among others. While these animals do offer an in vivo environment, their anatomy and physiology differs significantly from that of a human. The great expense and specialized facilities required limit their in-house use. Reproducibility may also be an issue as both inter- and intrasubject variability are difficult to control. Additional considerations include contention with the Animal Welfare Act, the significant expense associated with contracting regulated facilities and medical practitioners, and the risks related to handling biohazardous materials.
To get around these issues, developers tend to gravitate toward simpler and more accessible models such as cadavers and benchtop fixtures. Unfortunately, there tends to be an inverse relationship between the usefulness and complexity of the model employed. For example, cadaver tissues provide an accurate representation of anatomical geometry, but the required chemical preservation greatly alters the physical properties of the tissues. In addition, biological temperatures and flows cannot generally be simulated, subjects are difficult to source and maintain in useful quantities, and an educational institution must almost always be contracted (at considerable expense) to perform the study.
These factors drive early stage medical device developers to simple benchtop fixtures made (usually) in house by the developer. Unfortunately, these models are typically designed by individuals lacking an understanding of anatomy and physiology, and are usually fabricated from typical engineering materials such as metal, glass, and plastic. While an argument may be made that these models are better than nothing, they are certainly not representative of actual use conditions. Furthermore, engineers in general will agree that the quality of test data is dependent on the good logic behind the test protocol and the quality of the model employed. A poor model is therefore more likely to yield misleading data, and a design based at an early stage upon this data is more likely to require correction at a later stage in development.